Date: March 15th 2023

Time: 2PM-4PM

Location:

CRED, 21 Rue Valette, 75005 Paris.

First Presentation:

Nicola Borri

(LUISS University)

Nicola Borri

The Economics of Non-fungible Tokens

Abstract:

We construct a comprehensive dataset on a near universe of non-fungible token (NFT) transactions, create indices for the NFT market and its components, and analyze their properties. The NFT market return is significantly exposed to the cryptocurrency market return, but the majority of the NFT market variations remain unexplained. NFT market returns have low exposures to other cryptocurrency factors and factors from traditional asset markets. In the time-series, volatility and the NFT valuation ratio significantly predict NFT market returns. In the cross-section, NFT returns exhibit size and return reversal effects.


Second Presentation:

Francesco Zola

(Vicomtech)

Francesco Zola

Machine Learning and Generative Networks for improving Entity behavioural analysis in the crypto ecosystem

Abstract:

An escalation of illegal activities and the goal of improving the network’s resilience to cyber-attacks have led researchers and Law Enforcement Agencies (LEAs) to investigate how to reduce anonymity within the Bitcoin blockchain network. Anonymity can be decreased through Bitcoin entity classification, which aims to detect and classify entities’ behavioural patterns within the network. However, this classification problem - typically addressed via supervised machine learning approaches - strongly depends on the initial labelled Bitcoin dataset, which allows the definition of singular entity behaviour. Existing approaches for addressing the class imbalance problem can be improved considering generative adversarial networks (GANs) that can boost data diversity. However, despite their potential, it is impossible to find a unique GAN solution that works for every scenario. Therefore, one aim of this investigation is to study which type of GAN architecture can be used to generate synthetic Bitcoin behaviour, which architecture generates the most “valuable” information, and how the training time affects classification results.